ML19289F884

Submits Notes for 790517 Presentation to Commissioners Re Interim Operational Sequence of Events

ML19289F884
Person / Time
Site:	Crane
Issue date:	05/16/1979
From:	NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION I)
To:
References
OSP-790516, NUDOCS 7906210019
Download: ML19289F884 (22)

Text

.

s-5/16/79

~

NOTES FOR PRESENTATION TO COMMISSIONERS MAY 17, 1979 INTRODUCTION Before presenting the highlights of this interim operational sequence of events, I want $ to describYtheorganizationoftheteamwhichis conducting the investigation and to briefly summarize for you the primary sources of information which we utilize in the conduct of the investigation.

( Fi rst a..c.

%n.. ~

"fhe team which is conducting the investigation into the operational aspects hd of the accident consists of one supervisor,3five inspectors who have been drawn from regions, headquarters, and the Performance Appraisal Branch.

Aiding the inspectors are two investigators drawn from the regions an'd five additional investigators drawn from the Office of Inspector & Auditor. The seven investigators are shared between the two teams that make up the Three Mile Island-2 investigation.

?mw The primary souru.s of information used by the investigation team to establish the sequence of events, as well as the actions performed by the licensee and his staff, include:

(1) interviews conducted with the individua'h who were

=~

involved in some aspect of the accident.

These interviews are taped) 4

' transcriptions are prepared, and subsequently used to compare t. e information obtained from the individuals.

In addition to these taped interviews, members of the team hold extensive discussions with int.viduals for the purpose of determining how components and systems function or the reasons behind particular operator actions so that the formal interviews whic are recorded can be I'

more specific;

-241 12f,<

7 9 0 6 210 0 l%

4

-g-ck A._ %&'

T had-

$~t~. { 3 (LLLCg

-b Wh coudLd -ih RWA s %

Y adia4hwA y nd SkLo MAL w cA w y u.afk a

t; h. % w oe.4 M k tu14 >A p M 3 h q q L to Q X L

-e J h.

Ih rh%sa Ep qMx 3

&& wJ A & QQ L}'c/ i Yu au.

sfcknuZ u w>bd -fu +tu. y W, S "

VuAw 3

W WAualv an L urdg && ax hM

/dh6 1 7 f_./u_.,.

lA as +

mm Le.st/wuJs'. W/= ufa%L1 c

4 b,e.gi? 1 6. eJL L A r y &

atet

&M ry&C&ww4

~

i

'/

(2) recordings of telephone conversations wfaich were made'Af telephone calls between the licensee during the course of the incident and the Regional I Incident Center and the.:E Headquarters 1.1 Nenter: (3) strio

% wsqt wa$/auLapfM.'

24l 126 h b wl, e

r

. charts remcved from the various plant recorders for the period of time of the incident; (4) data obtained by de-logging of the reactimeter, which is a system which records on tape pre-selected plant parameters and is used by the licensee a

for interpreting plant data following transients; (5) logbooks maintained by individuals in the control room and elsewhere for the purpose of recording data during the course of the accident; (6) reference material on various system design details, such as might be contained in the Final Safety Ana'vsis Report, in the system descriptions prepared by the architect / engineer, in the training manuals provided to the operating staff, or other such reference material; (7) the output from the alarm printer, which is one of the functions of the plant computer system--with regard to this last item, some mention has been made of the fact that for approximately one and one-half hours covering the period of 73 min after the start of the accident, to 167 min after the start, the alarm printer was unavailable.

Based on our interviews with plant staff present in the control room at the time, the alarm printer was shutdown because the mechanism which feeds the paper through the typewriter had jammed, and the paper was no lenger advancing.

This resulted in the typewriter printing over the same line and virtually cutting the paper in half.

When time was available, an operator corrected the paper feed problem and put the unit back into service.

During the course of our investigatin we found evidence of other paper feed type problems, which would show up by misaligned typing which leads us to helieve that paper feed problems on that unit were not uncommon.

We do not believe that there is substantial signific.ance to the loss of this alarm printer information for the period involved, either to the operators during the course of the incident, or for cur purposes in investigating the events of the incident.

During a period of. time wher there is a large rate of ;1 arms 24i 177

~

I

' occurring in the control room and being typed on the printer, it is common for the printer to run substantially behind real time.

As a consequence of 5

this, the operators rarely use the alarm printer as an. operational tool during the course of any major evolution.

From our review standpoint, the loss of the alarm printer informatica causes us some incon tenience in identifying the specific times at which certain events took place, b;t since it is not the only source W +Fc,t#

7 of information, it will not prevent us from determining the events that took place during that period of time.

[4 ye.g.,

0 t.d W WW

,,tJaAfr~ M (Y"

's awaAr t m

M % s] -

dh}y

^

Q w

dyO) cod Ac s,jn L w~

C ecn &

.w W/

"o uu cM N, U * %

' w ds.kn Q u

~

/

A.k. Wf '~ ~ f k "7f u nX() c h d47.L 2wll, 241 123

~.

\\,u d g W. M u 4E %dw

• Mtw A

% vwt

- wC he p HIGHLIGHTS OF SEQUENCE OF EVENTS h

ELAPSED TIME EVENT / CONDITION Prior to Turbine Trip Reactor C3clant System Leakace i

Prior to the turbine trip, the licensee has suspected that either the EMOV and/or the pressurizer safety valves were leaking and the licensee was prepared to make a containment entry into the Unit 2 containment after any trip of the unit for the purpose of making further temperature measurements to determine which valve was leaking.

This valve leakage was the orcbable cause of there being a dif ference in baron concentration between the reactor coolant system and the pressurizer, which had led the licensee to establish a continuous spray in the pressurizer, which is Mw a standard technique for Wg # reactor coolant system and pressurizer boron concentrations ul inte -[ m

resamt.

The leakage which existed from these valves was within the limits permitted in the Technical Specifications and such valve leakage is not unusual for safety or relief valves in pressurized water reacters.

ELAPSED TIME EVENT / CONDITION Since the relief lines from all three valves join together to a ccanon line which goes to the reactor coolant drain tank, it is not possible to uniquely identify which valve was leaking from the temperature alarms provided on the line printer.

This is becausc any steam released to the common header results in raising the temperature indicated for all three valves.

The significance of the leaking valve is not clear at this point.

It is included in the operational sequence of events only as an indication that there may have been some difficulty with one or more of those valves which could have led to the failure of. the EMOV to re-seat following its opening after the turbine trip, t

Resin Blockage Difficulties were being experienced by the operating staff in transferring an isolated condensates system polisher spent resins to the re-generation receiving tank.

This transfer procedure uses. station air (a compressed a&

air system) to fluff the resin 43 de-mineralized water to transfer the resin between tanks.

Based on interviews, a resin blockace developed in the 241 130 I

I f ELAPSED TIME EVENT / CONDITION transfer line, and with the stoppage of flow, the water pressure in the polisher tank increased.

The plant operators have hypothesized that water pressure may have exceeded the air pressure which was being used to fluff the resin bed, resulting in a forcing of water into the air system.

They further hypothesize that the water made its way to the polisher isolation valve controls, causing them to drift toward the shut position.

Shutting of those valves would have resulted in reducing water flow to the condensate booster pumps and it may be that the condensate booster pumps were the ones which tripped first, based on the control logic of these pumps.

A M

~

241 131

7 Min. 43 Sec.

Reactor building sump ' pump A turns on.

p res uma bly ' i ni ti a ted ' au toma t i cal ly. due to level increasing in the reactor building sump, and based on design infonmation on the pump, should have been pumping approximately 140 gpm from the reactor building sump to the miscellaneous waste holdup tank via the normally open -reactor building isolation valves.

However, a later review of tank level records show that the miscellaneous waste holdup tank did not change level substantially during the incident and there is some reason to believe, based on interviews with personnel, that the sump pump discharge had been aligned to the auxiliary building sump tank.

This is a tank with a capacity of approximately 7000 gallons.

It also appears frcm interviews that this tank had a failed 241 132 e

ELAPSED TIME EVENT / CONDITION l'upturc disc which was scheduled to be repaired later.

This failed rupture disc did not fail during this incident, but apparently had failed at some prior time.

It should also be noted that the isciation valves on this line between the building sump pumps and either the miscellaneous waste tank or the auxiliary sump tank do not close on the safety signals which initiates high pressure injection.

These close only when a pressure of 4 psi is reached in the reactor building.

At lp min,19 sec, approxi-mately 156 seconds later, the second reactor building sump oump starts pumping into the same discharge line.

Based on design information both pumps should be discharging water from the reactor building to the auxiliary building at approximately 280 gpm.

The second building sump pump starts, based on a level rising in the sump which indi$tes that the first pump A

does not have enough pump capacity.

Both of these pumps continue to operate until approximately 38 min af ter the start of the incident. At that point, both sump pumps were turned off by an auxiliary operator.

The two pumps had operated min for 31 and 28/respectively pumping approximately 241 133

-g-ELAPSED TIME EVENT / CONDITION 8000 gallons of water to the auxiliary building.

It should be noted that shut i

down of the pumps does not result in isolation.

of the pump discharge line.

Again, this occurs only when a nigh pressure in the reactor building of 4 psi initiates isolation of the building.

It should also be noted that the auxi.iary building sump pump tank is located at an elevation in the auxiliary building, which might permit siphoning of tt.e contents of the reactor building over to the auxiliary building waste tank.

Moreover, this siphoning /h)$

WW could be acchd due to any pressure which would build up within the reactor building as a result of continued steam discharge.

We have cevieur' not completed the acetys+s to demonstrate that this was a hW path for water leakage from the reactor building the auxiliary building.

Mcwever, should it prove to be a primary source of the flooding which occurred in the auxiliary building, this line would not have been isolated until 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> and 17 min af ter the start of the incident.

This possibility is under investi, gation and is not considered to be an established fact at this time.

241 134

r $ ct r 4 g.~ u _ m iWaadAhve-e

~...-- m c.

RE

..7Crt"

~

n rwr.M SCRIPTION_

Operator considers 1.

' finds OTSG level at 10 inches en the Startup Range.

Operator verifies Emergency 2.

M[1cvel reans OTSG is dry" per his training.

)W1A Operator finds and 3.

feedwater Pumps are running and examines valve lineup.

Position indicating lights on 128 were 4.

announced that EF-V12A & 128 are shut.

.,.obscuredty a caution tag hanging from another valve controller.

Position indicting iights for 12A may have been obscured by operator s body as he leaned over panel.

Operator drives valves open, resulting in dry OTSG being fed with relatively cool Hot and cold leg temperatures drop.

RCS pressure, now under control of loop water.

-~ saturation conditions, follows accordingly.

A routine, scheduled surveillance test was performed on the A & B electric emergency feedwater pumps on March 26, 1979 at approximately 10:00 a.m.

Implementation of this surveillance test procedure results in closure of both the 12A and 12B valves, regardless of which pump is being tested.

The procedure calls for reopening of the valves, along with insuring the proper status of at least three other valves.

The procedure for the electric driven emergency feedwater pumps is insuffic-iently specific to provide documentation of valve opening in that the procedure does not require individual signoffs frr each valve; rather, the procedural requirement is in sentence form with one signoff signifying proper positioning of the valves.

The investigation thus far has not found any evidence of willful closure of the y

valves over the period of 3/26/79 to 3/28/79 TL -t'g._4.-_%

~ #

f c La.+c.d (On 3/23/79, an tiRR Operator Licensing Branch (OLB) Examiner conducted exams and 1.

"wgikcq thru" the EFW System.

The valves were open on that day. )

~.:n==c'

' Rapid rise in OTSG pressure observed indicating feed flow to generators.

Confirmed 1.

f'5 by EFW pump discharge pressure decreasing and "hamering" and " crackling" heard 2.

9 from the Vibration and Loose Parts Monitor Speaker aligned to listen to "A" 0TSG.

• • < %::>aUiM a",*i& ' "

.-w- _m:>.m sm,w --

J-

^

f

-9A M A -n am w

p >'

n t, rm_ cr

'pm( ' ' :'+/ h -

1.

Sac. cad-Re:wu r La G u i o9 :_.p -pep--(-W9L P 28) - r ' wk t ' gup%.;ca u to r DufM?rrg 1.

Susp tcL-the-Audliary'-Bu a i a n.9 e u a C gpm.

ide.. o n.Jp.

tb te : O c=d "nct;i L ildo.3

'u... g puu.y narca vuoco oo,.a.o3 I x Jd.

l.

Nke-tip pc'ip-Y$-tT rp a, ;id i;$taded

'"4' "r'""'*

E.p2 qu. o n.

Pm.or Sui 4dir+t W an:ww

$6.5_ f-t).

1.

p r-T on ro m Opa cor re

..p ep

~

1.

e

...c-7 1me this int.

741 135 8

T a T rd M Ml W y$N % g f E Nfl.}E W Q q r,; g s w n w c PWT' m7m

...: 5 A m-

~.

ia

~W8WG$:

?

QV Q

e.mme.w. w. w. &mhga[m.y

,a,l s

h,. w: o u w..~ m,n p.

w m

a f

h hkI b

b bm$e

.s n o s e n w;e[a$! m m m%q$$

g@u@n!%#u@hgM@adg!#p@wMM 4#!WMMMy gy 5

E dM er WQJ%f igg WMp dgii IIIbbMb !$hp$$kNNhhbgbb,4.aw.hffyhd'hhh A

w.m.

w

.w y

s

$2dffshh i

15 Min.

Reactor Coolant Drain Tank Rupture Disc Blows The opening of the reactor coolant rupture disc at 15 min into the incident has been reported p revi ously. NFy investigation shows that the RCDT vent lines isolate automatically at 10 psig on rising pressure, but in order to protect the RCDT from tank collapse in the event of a cooldown, that vent line valve reopens at approximately 6 ' psig.

Reopsning of that vent line valve essentially opens the reactor coolant drain tank to the waste gas system.

The waste gas system header is maintained at 2 psig and as long as isolation of.the reactor building does not occur, it is feasible that during the period of time that the reactor building pressure was in excess of 2 psi, and isolation had not occurred, there could have a,ftIAN% afat6 ppd % -

been a transport ofs h the reactor building to the waste gas system to th; ;sur;;md in the auxiliary building.

py (c

- {r N Y

A Qasley 3

, ELAPSED TIME EVENT / CONDITION 74 Min.

Stopoace of Reactor Co'lant Pumos o

100 Min.

Shortly after the start of the incident, and up until the time the reactor coolant pumps were secured, the operators were receiving multiple alamrs and indications from the reactor coolant pumps.

The included high vibrational alams, full speed alams, which would suggest vapor binding or a shaft failure, and low motor current alarms, which would indicate shaft failure, coupling failure, or vapor binding of the pump.

Interviews withthe operators disclosed that they feared a seal failure LOCA and decided to go on natural circulation rather than risk destruction of the reactor coolant pump seals.

During the period when these pumps were shut down, the operating staff did not recognize that they did in fact have a LOCA frcm EMOV failure.

V

-! DAW es udix4k.setaa/ &

4d6~ %A

,, a yaf e~ uaa a a dm (yh ~/$SYsOf tcf/Gt 1HG 2

( % 2{5 u d 9 0 / / m )

hunNWE 241 137

=

~ A ELAPSED TIME EVENT / CONDITION 75 Min.

Low Boron Concentration in Reactor Coolant 90 Min.

System Samoles At both of these times the operators in the control roca are informed that the boron samples were showing low boron concentrations.

The first sample indicated 700 ppm boron, which is substantially belcw the boron con-centration of 1025 ppm which existed in the RCS prior to the trip.

Based on the first low sample, the results were questioned by a technician and he took steps to get an additional sample to check his first results.

The 90 min sample was reported to be in the range of 400-500 parts /million baron, With a radioactivity concentration approximately ten times that which was in the reactor coolant system before the start of the incident.

As of this time, the investigation has found no explanation for these low ' values of baron concentration, and is an item which will continue to be pursued.

During these periods in which the low boron concentration results were reported, the operators had also noted varying source range instrumentation levels which were abnormal and which caused the operators to " emergency 2

borate" at approximately 90 min and again at 100 mi, 241 138

NU

. 2 LAPSED TIME EVENT / CONDITION The interviews with the operators disclosed that they were concerned because of the source range instrumentation behavicur that they had experienced a re-start of the reactor despite the fully inserted rods.

It appears that the count rate behaviour coincident with the reported low boron concentrations served to distract the operators to some degree with concerns over the extent to which the reactor 'was fully shut dcwn.'

i s

I

~

2 Hrs. 54 Min.

Additional Reactor Coolant Pumo Ooeratior At this time, the operating staff was unconvinced that they had successfully established natural circulation in the reactor caolant system.fe'Iawing the shut down at 100 min of the last of the reactor coolant pumps.

Reactor coolant pump 2B was put into operation and almost simultaneous:

with its operation virtually all the radiation monitors which annunciate in the reactor control room al a rmed.

A review of the plant strip charts and other data associated with the once-through steam generators suggest that that flow was successfully produced, but within 19 min the operators started receiving alarms and indications such as low motor ampherage and full speed alarms, similar to those received when they decided to shut down the pumps originally.

19 min later the pump was shut down. Additionally, at about 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> and 8 min after theiincident, at least two pumps were operated briefly again, but all gave no-flow-type running currents, convincing the staff at that time that there was steam in the lcops, and that pump operation would not be effective.

s 24l 140

m

• pr;.si 2y 7 "-e No RCP Operation While System at High Pressure 7

(5 1/2 to 7 1/2 hours) r During this period, the operations staff was reluctant to start a reactor coolant pump for fear of vibration Mcketd seal failyre and a resultant seal failure LOCA.

They recognized that they had-steam bubbles in both reactor coolant system loops and thus would be operating the pumps in a steam environment.

Interviews disclosed that at this time perscnnel I/f believed that the core was covered and as each option for further actions to be taken was reviewed it is : reported that the potential for uncovering the core was part of that review process.

Moreover, trie presence of the bubbles in both loops led them to conclude that the steam generators would not be effective at this time as a means _of_._

removing decay heat.

Thus, they directed th6ir attention to developing other methods of establishing an adequate heat sink for the reactor.

O 5

241 141

7 Hrs. 30 Min.

Decisi0n to De-pressurize to Utilize Core Flood Tanks Based on the interviews covering this period, the operators indicated that they were aware they were having trouble maintaining a sufficient number of pressurizer h~ eaters and they were concerned that the blocked valve en the EMOV might fail in an open position which would eliminate their ability to isolate the EMOV should it malfuncticn again.

Furthe rmo re,

I e

4 9

e 4

0 4

24I i42

_4_

ELAPSED TIME EVEtiT/CCNDITI0t1 as mentioned earlier regarding the attempts to run the reactor coolant pumps, they were

=

still concerned about the steam bubbles in t'he ility of the stean generators to be used as a source of heat removal.

It was then decided to de-pressurize the system f to approximately 600 psi where the core flood tanks would be available to add their water in the event that they did not have sufficient water in the core.

Based on our interviews it appears that the decision to go on to the decay heat removal system at a pressure of 400 lbs. was made after-the de-pressurization had started, and was not an initial consideration that entered into the de-pressurization.

When they reached the core flood tank cover gas pressure of 600 lbs., th'ey experienced only a small reduction in core flood tank level as they passed below the CFT blanket pressure. At approxircately 9 hours1.041667e-4 days <br />0.0025 hours <br />1.488095e-5 weeks <br />3.4245e-6 months <br /> a,nd 15 min tu into the accident, they attributed the ability b-to drop below 450 psig as a sign that they had reached saturation pressure of the loops and would be unable to dc-pressurize 241 143

. ELAPSED TIME EVENT / CONDITION further witho-some mechanism for cooling thea.e loops.

During the period of 10 to 11 1/2 hours after the start of the incident, a series of indications on the alam printer of oscillating core flood tank levels were received.

Investigation has not been able to establish the reasons for this oscilatory

&(

behaviour.

However, it appears that the 4

[

1.

failure of the core flood tanks to discharge

{lb their contents nr ut to convincing the f 'f

[0 Q

staff that the core was covered and that they M,?

had been able to achieve some degree of stable J ~ht>'

conditions in the reactor coolant system.

N 11 Hrs.12 Min. to No Heat Removal From Steam Generators 13 Hrs. 23 Min.

At approximately 11 hours1.273148e-4 days <br />0.00306 hours <br />1.818783e-5 weeks <br />4.1855e-6 months <br />,12 min, due to concern over the possibility of atrcspheric contamination by using the atmospheric steam dumps, use of the atmospheric steam dumps was

• teminated.

Prior to this point, the State had indicated some concerns of the possibility of contamination due to continued use of the atmcspheric steam dump.

'When use of the atmospheric steam dump was terminated, it was not possible for the licensee to go to the steam dump to the condenser because they had

=

lost condenser vacuum.

They had lost the 241 144

, ELAPSED TIME EVENT / CONDITION steam necessary to maintain the seals on the turbine; thus, they had lost condenser vacuum 4

and would have to re-establish it.

During this period of time, coolant for the core relied on high pressure injection ficw and the existence of the core flood tanks riding on the reactor coolant ' system.

The lice.1see was also reluctant to initiate further blowdown with a corresponding increase of high pressure injection flow due to the potential need to go onto recirculation once the SWST was empty and use dirty sump water to continue the process (entry at 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br />, 56 min).

At approximately 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br />, 30 min the licensee decided to establish a vacumm in the condenser and to start steaming the A steam generator to the condenser and remove heat through natural circulation.

It was at about this time that the licensee believed that the A loop had Qcne d

u Igsolid again under the action of the high pressure injection.

Vacucm in the condenserrand the ability to steam the condenser was re-established by approximately 13 hours1.50463e-4 days <br />0.00361 hours <br />2.149471e-5 weeks <br />4.9465e-6 months <br />, 23 min, but before any substantial steaming was started the licensee had also decided to initiate re-pressurizat-and to attempt operation of the reactor coolant pump.

Based on the data reviewed thus far, 241 145

~

EVENT / CONDITION ELAPSED TItiE_

it it not clear that they successfully established natural ctculation cooling when they began steaming of the A steam generator.

These remarks were prepared for the purpose of pointing out those items in the interim operational sequence of events which expand information which had been provided to your previously.

• 4 5

241 146